项目简介
Hiring Agent 解析简历 PDF，利用大模型提取结构化数据，结合 GitHub 信号，输出公平可解释的评分。支持完全本地运行（Ollama）或使用 Google Gemini。
README 预览
# Hiring Agent\n\nResume-to-Score pipeline that extracts structured data from PDFs, enriches with GitHub signals, and outputs a fair, explainable evaluation.\n\n\n \n \n \n \n \n \n \n \n \n\n\n---\n\n## Contents\n\n- [Overview](#overview)\n- [Architecture](#architecture)\n- [Installation and Setup](#installation-and-setup)\n - [Prerequisites](#prerequisites)\n - [Quick setup with pip](#quick-setup-with-pip)\n - [Ollama models](#ollama-models)\n- [Configuration](#configuration)\n- [How it works](#how-it-works)\n- [CLI usage](#cli-usage)\n- [Directory layout](#directory-layout)\n- [Provider details](#provider-details)\n- [Contributing](#contributing)\n- [License](#license)\n\n---\n\n## Overview\n\nHiring Agent parses a resume PDF to Markdown, extracts sectioned JSON using a local or hosted LLM, augments the data with GitHub profile and repository signals, then produces an objective evaluation with category scores, evidence, bonus points, and deductions. You can run fully local with Ollama or use Google Gemini.\n\n---\n\n## Architecture\n\n\n\n\n\n**Flow**\n\n1. `pymupdf_rag.py` converts PDF pages to Markdown-like text.\n2. `pdf.py` calls the LLM per section using Jinja templates under `prompts/templates`.\n3. `github.py` fetches profile and repos, classifies projects, and asks the LLM to select the top 7.\n4. `evaluator.py` runs a strict-scored evaluation with fairness constraints.\n5. `score.py` orchestrates everything end to end and writes CSV when development mode is on.\n\n\n\n\n**Key modules**\n\n- `models.py`\n Pydantic schemas and LLM provider interfaces.\n\n- `llm_utils.py`\n Provider initialization and response cleanup.\n\n- `transform.py`\n Normalization from loose LLM JSON to JSON Resume style.\n\n- `prompts/`\n All Jinja templates for extraction and scoring.\n\n\n\n\n\n---\n\n## Installation and Setup\n\n### Prerequisites\n\n- **Python 3.11+**\n\n The repository pins `.python-version` to 3.11.13.\n\n- **One LLM backend** (either of